Substitution of l-Tryptophan by α-Methyl-l-Tryptophan in 177Lu-RM2 Results in 177Lu-AMTG, a High-Affinity Gastrin-Releasing Peptide Receptor Ligand with Improved In Vivo Stability

Visual Abstract

Radi oligand therapy has emerged as a powerful alternative to conventional treatment options in oncology. This emergence can be attributed mainly to the success of DOTATOC-and DOTA-TATE-based theranostics in the case of neuroendocrine tumors and to prostate-specific membrane antigen (PSMA)-targeted inhibitors in the case of prostate cancer (1,2). In view of the overexpression of the gastrin-releasing peptide receptor (GRPR, bombesin-2 receptor) at a high density and frequency already in early stages of prostate cancer (5,000 disintegrations/min [dpm]/mg of tissue, with .2,000 dpm/mg being considered clinically relevant) and breast cancer (10,000 dpm/mg), GRPR has been identified as a promising target for both cancer types (3,4).
As we hypothesized that the use of statine (i.e., Sta 13 ) at the C terminus of RM2 and its derivatives would result in sufficient metabolic stabilization at this part of the molecule, we concluded that further improvements might be possible by stabilizing the Gln 7 -Trp 8 sequence. For this purpose, we substituted a-methyl-L-tryptophan (a-Me-L-Trp) for L-Trp 8 in 177 Lu-RM2 and its DOTAGA analog ( Fig. 1) and evaluated these novel compounds alongside the potent reference ligands 177 Lu-RM2 and 177 Lu-NeoBOMB1. The comparative preclinical evaluation comprises affinity studies (half-maximal inhibitory concentration, or IC 50 ) on PC-3 and T-47D cells, quantification of receptor-mediated internalization on PC-3 cells, determination of lipophilicity at pH 7.4 (logD 7.4 ), investigations of stability against peptidases in vitro in human plasma and in vivo in plasma and urine of mice, and biodistribution studies on PC-3 tumor-bearing mice.
Both nat Lu and 177 Lu labeling was according to a modified procedure (23). The radiolabeled reference, 3-125 I-D-Tyr 6 -MJ9 (Supplemental Figs. 1 and 2; supplemental materials are available at http://jnm. snmjournals.org), was prepared according to a reported procedure (24). A detailed description of the synthesis, labeling, and characterization of RM2 and its analogs is provided in Supplemental Figures 3-12.

In Vitro Experiments
A detailed description of all cell-based experiments is provided in the supplemental materials.
Determination of Lipophilicity (n-Octanol-Phosphate-Buffered Saline Distribution Coefficient, logD 7.4 ). Approximately 1 MBq of the 177 Lu-labeled compound was dissolved in 1 mL of phosphate-buffered saline (pH 7.4) and n-octanol (v/v 5 1/1). After stirring for 3 min at room temperature in a vortex mixer and subsequent centrifugation at 9,000 rpm for 5 min (Biofuge 15; Heraeus Sepatech GmbH), 200-mL aliquots of both layers were measured separately in a g-counter. The experiment was repeated at least 5 times.
In Vitro Stability Studies. Metabolic stability in vitro was determined through a procedure published by Linder et al. that was slightly modified (16). Immediately after labeling, 200 mL of human plasma were added and the mixture was incubated at 37 C for 72 6 2 h. Proteins were precipitated by treatment with ice-cold EtOH (150 mL) and ice-cold MeCN (450 mL), followed by centrifugation (13,000 rpm, 20 min). The supernatants were decanted and further centrifuged (13,000 rpm, 10 min) using a Costar Spin-X (Corning) centrifuge tube filter (0.45 mm). The filtrated plasma samples were analyzed using radio-RP-HPLC.

In Vivo Experiments
All animal experiments were conducted in accordance with general animal welfare regulations in Germany (German animal protection act, as amended on 18.05.2018, article 141 G v. 29.3.2017 I 626, approval ROB-55.2-2532.Vet_02-18-109) and the institutional guidelines for the care and use of animals.
For biodistribution studies, organs were removed, weighed, and measured in a g-counter (Perkin Elmer) after euthanasia at 24 h after injection.
Imaging studies were performed on a MILabs VECTor 4 smallanimal SPECT/PET/optical imaging/CT device (MILabs). Data were reconstructed using the MILabs Rec software (version 10.02) and a pixel-based algorithm (similarity-regulated ordered-subsets expectation maximization), followed by data analysis using PMOD software (version 4.0; PMOD Technologies LLC). Static images were recorded at t 5 1, 4, 8, 24, and 28 h after injection with an acquisition time of t 1 (45-60 min) using a high-energy general-purpose rat and mouse collimator and a stepwise multiplanar bed movement.

Synthesis and Radiolabeling
Uncomplexed ligands were synthesized via standard Fmoc-based solid-phase peptide synthesis, yielding 9%-15% of each labeling precursor after purification by RP-HPLC (chemical purity . 98%, determined by RP-HPLC at l 5 220 nm). Complexation of all ligands with a 2.5-fold excess of nat LuCl 3 resulted in quantitative yields. The remaining free Lu 31 did not affect the cell-based assay in a brief competition study (Supplemental Fig. 13); thus, purification before affinity studies was dispensed with. 125 I-iodination of D-Tyr 6 -MJ9 by means of the Iodo-Gen (Pierce Biotechnology, Inc.) method resulted in 3-125 I-D-Tyr 6 -MJ9 with radiochemical yields of 33%-48% and radiochemical purities of more than 98% after RP-HPLC purification. 177 Lu labeling of all compounds was performed manually, each resulting in quantitative radiochemical yields, radiochemical purities of more than 98%, and molar activities of 40 6 10 GBq/ mmol. All 177 Lu-labeled ligands were used without further purification.
Biodistribution studies on PC-3 tumor-bearing mice were performed at 24 h after injection (Table 1). 177 Lu-RM2 and its derivatives showed low activity levels in most organs at 24 h after injection, indicating a rapid clearance from nontumor tissue, as is especially important for blood and GRPR-positive organs such as pancreas and intestine. 177 Lu-NeoBOMB1 showed increased activity levels in several nontumor organs at 24 h after injection, particularly in lung, liver, spleen, pancreas, intestine, and adrenals. Bone uptake was slightly enhanced for 177 Lu-RM2, a finding that was attributed to incomplete labeling (radiochemical yield, 95%; chromatogram not shown) and thus free 177 LuCl 3 . Tumor retention was comparable for all compounds except 177 Lu-AMTG, which exhibited distinctly increased values at 24 h after injection (7.2-8.5 vs.

percentage injected dose per gram; %ID/g). Not surprisingly, 177
Lu-AMTG showed the highest tumor-to-background ratios at 24 h after injection. The tumor-to-blood ratio of 177 Lu-AMTG (2,702 6 321) was almost 4 times higher than that of 177 Lu-RM2 and 177 Lu-AMTG2 and approximately 15 times higher than that of 177 Lu-Neo-BOMB1 (Supplemental Table 2).
Small-animal SPECT/CT studies with 177 Lu-RM2 and 177 Lu-AMTG at 1, 4, 8, 24, and 28 h after injection in PC-3 tumorbearing mice revealed low background activity for both tracers at 4 h and beyond and, for 177 Lu-AMTG, considerably higher activity in both tumor and pancreas (Fig. 3). For both tracers, specificity of tumor uptake was confirmed via competition experiments with an excess of nat Lu-RM2 (Table 1; Supplemental Fig. 17).

DISCUSSION
With regard to radioligand therapy, the 2 most promising GRPR ligands, 68 Ga-RM2 and 68 Ga-NeoBOMB1, present some disadvantages: 68 Ga-RM2 suffers from rapid metabolic degradation (17), which is why tumor accumulation and tumor dose for 177 Lu-RM2 are likely limited as well-especially important in the context of radioligand therapy. In contrast, 177 Lu-NeoBOMB1, which exhibits a higher metabolic stability in vivo, shows enhanced activity retention in tumor tissue but also in blood (19). This characteristic results in unfavorable dosimetry and higher doses to the red bone marrow (25). With the aim of retaining the excellent pharmacokinetics of RM2, we substituted the unnatural amino acid a-Me-L-Trp for the metabolically less stable Gln 7 -Trp 8 sequence of 177 Lu-RM2 and its DOTAGA analog and compared these new ligands with 177 Lu-RM2 and 177 Lu-NeoBOMB1 as references. Synthesis was easily accessible via solid-phase peptide synthesis, and complexation with nat Lu or 177 Lu proceeded quantitatively.   All 4 compounds contain a similar pharmacophore typical of bombesin analogs, resulting in high affinities that were in the range of IC 50 values reported for nat In-RM2 (9.3 nM), several nat Ga-RM26 derivatives (2.3-10.0 nM), nat Ga-NeoBOMB1 (2.5 nM), and SB3 (3.5 nM) (18,19,21,26). Apart from nat Lu-AMTG2, higher cellular uptake of 3-125 I-D-Tyr 6 -MJ9, as well as slightly higher IC 50 values on PC-3 cells than on T-47D cells, was observed (Supplemental Figs. 18 and 19). These findings were attributed to an increased number of receptors on PC-3 cells.
We could show that a-Me-L-Trp-for-L-Trp 8 and DOTAGAfor-DOTA substitution had only minimal impact on GRPR affinity, lipophilicity (logD 7.4 ), and receptor-mediated internalization, demonstrating that these modifications might allow the in vivo kinetics of 177 Lu-RM2 to be kept almost unaffected. In contrast, higher internalization levels and lipophilicity already indicate the in vivo limitations of 177 Lu-NeoBOMB1.
Besides retaining the favorable in vitro data of 177 Lu-RM2, the primary aim of this study was to chemically stabilize the Gln-Trp bond to potentially improve its longtime behavior in vivo. Comparative stability studies in vitro and in vivo, as well as the resulting biodistribution profiles, substantiated our working hypothesis of addressing the major metabolic instability at the Gln-Trp site in RM2 and other bombesinlike compounds. Both 177 Lu-AMTG and 177 Lu-AMTG2 exhibited equal or even increased amounts of intact compound in human plasma in vitro and in murine plasma and urine in vivo, in comparison with the 2 references. For 68 Ga-RM2 and 177 Lu-NeoBOMB1, the fraction of intact tracer in murine blood was reported to be 55% (15 min after injection) (27) and 90% (30 min after injection) (19), respectively, which are lower than the value we determined for 177 Lu-AMTG (30 min after injection).
Unlike Linder et al. in a stability study on 177 Lu-AMBA (16), we observed fewer metabolites for each ligand after incubation in human plasma (Supplemental Fig. 14), as can be explained by the C terminus modifications present in each of the 4 compounds tested in this study. Popp et al. observed 1 major and 2 minor metabolites for 68 Ga-RM2 in murine plasma at 15 min after injection (27), whereas we detected only 1 major and 1 minor metabolite for 177 Lu-RM2 at 30 min after injection. This difference could be due either to our analysis method or to the effect reported by Linder et al. that the minor metabolites can be further metabolized to yield the major metabolite, the longer the circulation in vivo takes place.
Not surprisingly, increased metabolic stability observed in human and murine plasma for 177 Lu-AMTG resulted in a 35% higher activity level in PC-3 tumor for 177 Lu-AMTG than for 177 Lu-RM2 at 24 h after injection (Fig. 4). Both 177 Lu-AMTG and 177 Lu-AMTG2 exhibited excellent clearance kinetics and thus low activity levels in nontumor organs, with the highest values obtained for the kidneys (1.2-1.9 %ID/g). Both compounds were cleared mostly intact (Supplemental Fig. 16), a finding that could favor 177 Lu-AMTG and 177 Lu-AMTG2 over 177 Lu-RM2 and 177 Lu-NeoBOMB1, as charged metabolites tend to be taken up by and retained in the kidneys. Most importantly, the activity concentration in the blood and in the GRPRpositive pancreas was low for all 177 Lu-RM2 analogs at 24 h after injection (,0.01 and ,1 %ID/g, respectively), which we considered another prerequisite for successful translation into humans.
In contrast, 177 Lu-NeoBOMB1 displayed enhanced activity in most nontumor organs and thus the lowest tumor-to-background ratios in most organs, especially in blood, liver, spleen, pancreas, and adrenals, a finding that was also observed by other groups (19,25). The biodistribution profiles confirmed our concerns about increased lipophilicity and internalization rates. It might be speculated that retention in the GRPR-positive pancreas could be caused by a partial agonistic behavior observed in our internalization study, since GRPR agonists such as PESIN or AMBA that exhibit internalization rates of more than 25% at 1 h in vitro typically show a slow clearance from the pancreas over time (28,29).
Although reduced internalization might have other causes, the high structural similarity of 177 Lu-AMTG/ 177 Lu-AMTG2 to the known GRPR antagonist 177 Lu-RM2 and the comparably low internalization pattern observed in our studies are strong indicators of antagonistic behavior by these new compounds. This assumption is further corroborated by rapid pancreatic clearance within 24 h after injection and the resulting favorable biodistribution profiles. Evidence was also provided by small-animal SPECT/CT scans with 177 Lu-RM2 and 177 Lu-AMTG over time, both of which showed high tumor retention and fast clearance from nontumor organs, even the GRPR-positive pancreas. It is noteworthy that clearance from pancreas and tumor was less rapid for 177 Lu-AMTG, confirming our hypothesis on increased metabolic stability in vivo generated by a simple modification at the Trp 8 site. Thus, it is not surprising that tumor-to-background ratios for 177 Lu-AMTG were highest in all organs, except for the tumor-tomuscle ratio (Fig. 5).
Regarding dose-limiting organs in the context of radioligand therapy, the excellent tumor-to-kidney and tumor-to-blood ratios make 177 Lu-AMTG a highly attractive alternative to 177 Lu-RM2 (30). In fact, 177 Lu-AMTG seems to synergistically combine the advantages of 177 Lu-RM2 and 177 Lu-NeoBOMB1 regarding pharmacokinetics and stability while simultaneous offering the best GRPR affinity, both on PC-3 cells and on T-47D cells. Thus, a clinical assessment (e.g., clinical phase I study) with 177 Lu-AMTG seems warranted.
In summary, we were able to successfully introduce an a-Me-L-Trp-for-L-Trp 8 substitution within the pharmacophore of 177 Lu-RM2 that not only resulted in a new tracer ( 177 Lu-AMTG) with comparable affinity, internalization, and lipophilicity but also resulted in considerably improved metabolic stability. Hence, improved tumor uptake and pharmacokinetics superior to those of the parent peptide, 177 Lu-RM2, or the second reference compound, 177 Lu-NeoBOMB1, were observed for 177 Lu-AMTG. A noteworthy finding is that improved metabolic stability was achieved without coadministration of peptidase inhibitors (21), such as phosphoramidon; this finding could facilitate clinical translation. It seems legitimate to conclude that other bombesin derivatives published in recent years, which have been modified at the N or C terminus but not at the unstable dipeptide sequence Gln 7 -Trp 8 (20)(21)(22), would also benefit from an a-Me-L-Trp-for-L-Trp 8 substitution. Nevertheless, studies on prostate and breast cancer patients have to be performed to show whether these promising preclinical results are reflected on a clinical level. CONCLUSION We could demonstrate that the new 177 Lu-RM2 derivative 177 Lu-AMTG offers better overall preclinical performance than 177 Lu-RM2 or 177 Lu-NeoBOMB1. On the basis of these results, a clinical translation of 177 Lu-AMTG is highly recommended to assess a potential improved therapeutic value for radioligand therapy of GRPR-expressing malignancies, such as prostate and breast cancer. In addition, we expect that substitution of L-amino acids by their corresponding a-alkyl-L-amino acid analogs could also be a valuable approach to stabilize the pharmacophore of other peptidic ligands that suffer from insufficient stability in vivo.

DISCLOSURE
A patent application has been filed on modified GRPR-targeted ligands, including AMTG, with Thomas G€ unther and Hans-J€ urgen Wester as inventors. Parts of this study were funded by the SFB 824 (DFG Sonderforschungsbereich 824, project Z [Hans-J€ urgen Wester]) from the Deutsche Forschungsgemeinschaft, Bonn, Germany. Hans-J€ urgen Wester is founder and shareholder of Scintomics GmbH, Munich, Germany. No other potential conflict of interest relevant to this article was reported.